Apache Ignite: test utilities - unit-testing

Does Apache Ignite provide any test utilities that one can use to easily spin up a multiple node cluster in tests for unit tests and integration tests?

There is no such thing as far as I know. It's easy enough to start multiple Ignite nodes in the same JVM without any additional tools.
To start multiple nodes in the same JVM you need to specify igniteInstanceName in configuration of each node.
To make all nodes see each other it's convenient to use TcpDiscoveryVmIpFinder without any addresses in it and with shared flag set to true. It will make all nodes that use the same instance of such IP finder see each other.
The following code can be used to create node configuration:
static final TcpDiscoveryIpFinder IP_FINDER = new TcpDiscoveryVmIpFinder(true);
IgniteConfiguration getConfiguration(String instanceName) {
IgniteConfiguration igniteConfig = new IgniteConfiguration();
igniteConfig.setIgniteInstanceName(instanceName);
TcpDiscoverySpi discoverySpi = new TcpDiscoverySpi();
discoverySpi.setIpFinder(IP_FINDER);
igniteConfig.setDiscoverySpi(discoverySpi);
return igniteConfig;
}

Related

Cannot build Corda (release/os/4.6) on Ubuntu 20.04 due to failing tests (CryptoUtilsTest & NodeH2SecurityTests)

I am trying to build Corda on Ubuntu 20.04. I have the latest sources from the git repo (release/os/4.6) and I run ./gradlew build in the main folder. However the build fails during two tests (see the detail description below). Is there something that I'm missing? Are there some special flags that I should use for building Corda?
First, the test test default SecureRandom uses platformSecureRandom fails at the last assert, i.e.,
// in file net/corda/core/crypto/CryptoUtilsTest.kt
fun `test default SecureRandom uses platformSecureRandom`() {
// [...]
// Remove Corda Provider again and add it as the first Provider entry.
Security.removeProvider(CordaSecurityProvider.PROVIDER_NAME)
Security.insertProviderAt(CordaSecurityProvider(), 1) // This is base-1.
val secureRandomRegisteredFirstCordaProvider = SecureRandom()
assertEquals(PlatformSecureRandomService.algorithm, secureRandomRegisteredFirstCordaProvider.algorithm)
}
The reason for the failed test is Expected <CordaPRNG>, actual <SHA1PRNG>..
For some reason, the test is successful if before inserting the provider, I call the getServices() method, i.e.,
val provider = CordaSecurityProvider()
provider.getServices()
Security.insertProviderAt(provider, 1) // This is base-1.
I also tried to get the service SecureRandom.CordaPRNG directly from the provider and it works, i.e,
println(provider.getService("SecureRandom", "CordaPRNG"))
prints out Corda: SecureRandom.CordaPRNG -> javaClass
Second, the test h2 server on the host IP requires non-default database password fails since it expects a CouldNotCreateDataSourceException, but it gets a NullPointerException instead, i.e.,
// in file net/corda/node/internal/NodeH2SecurityTests.kt
fun `h2 server on the host IP requires non-default database password`() {
// [...]
address = NetworkHostAndPort(InetAddress.getLocalHost().hostAddress, 1080)
val node = MockNode()
val exception = assertFailsWith(CouldNotCreateDataSourceException::class) {
node.startDb()
}
// [...]
}
The problem is that the address is 127.0.1.1:1080, which means that net/corda/node/internal/Node.kt::startDatabase() does not throw CouldNotCreateDataSourceException since the condition to enter the branch
if (!InetAddress.getByName(effectiveH2Settings.address.host).isLoopbackAddress
&& configuration.dataSourceProperties.getProperty("dataSource.password").isBlank()) {
throw CouldNotCreateDataSourceException()
}
is not satisfied. Instead it calls toString() on the parent of the path given by the DB name; the parent is null, and thus, it throws NullPointerException, i.e.,
val databaseName = databaseUrl.removePrefix(h2Prefix).substringBefore(';')
// databaseName=my_file
val baseDir = Paths.get(databaseName).parent.toString()
Unfortunately there't a LOT of reasons that building Corda from source might not work.
Here are my recommendations:
it could be a java issue, there's a docs page on the specific version of java 8 that you need to use, (latest java support is on the roadmap for corda 5 🔥) Here's the docs page on that https://docs.corda.net/docs/corda-os/4.5/getting-set-up.html
Like Alessandro said, you'll want to be aware that 4.6 isn't generally available yet so there may well be bugs and problems in the code until the release. In addition just take another look at the docs page on building corda (here: https://docs.corda.net/docs/corda-os/4.5/building-corda.html#debianubuntu-linux), it's mentioned to use 18.04 but not the latest linux, as there might be some random clib things that could get in the way there.
Good luck to you.

Distribute work stored in table to multiple processes

I have a database table where each row represents a work to be done. This table is filled up/receive work through a rest API. Apart from a rest-service taking up the work, I have another service which uses actors to process this work.
I need suggestions in distributing this work evenly across these workers. This work is not one time, it is kind of done at an interval until user deletes that.
Therefore I need a mechanism where
The work as it comes is distributed evenly.
If the second service(work consumer) fails it can again boot up with all the records in table and distribute the work again.
Each actor represents one row of the work table.
class WorkActor(workId: String)(implicit system: ActorSystem, materializer: ActorMaterializer) extends Actor {
// read the record from table or whereever you want to read
override def preStart(): Unit = {
logger.info("WorkActor start ===> " + self)
}
override def receive: Receive = {
case _ => {}
}
}
Create an Akka cluster sharding region to dispatch the request from rest api to corresponding actor. Calling startShardingRegion function to return an actorRef. Then you could send the message to this sharding actorRef by rest API, and then corresponding will help you handle the message.
final case class CommandEnvelope(id: String, payload: Any)
def startShardingRegion(role: String)(implicit system: ActorSystem) = {
ClusterSharding(system).start(
typeName = role,
entityProps = Props(classOf[WorkActor]),
settings = ClusterShardingSettings(system),
extractEntityId = ClusterConfig.extractEntityId,
extractShardId = ClusterConfig.extractShardId
)
}
// sharding key
object ClusterConfig {
private val numberOfShards = 100
val extractEntityId: ShardRegion.ExtractEntityId = {
case CommandEnvelope(id, payload) => (id, payload)
}
val extractShardId: ShardRegion.ExtractShardId = {
case CommandEnvelope(id, _) => (id.hashCode % numberOfShards).toString
case ShardRegion.StartEntity(id) => (id.hashCode % numberOfShards).toString
}
}
Read or recover the data from preStart function in the actor. There are many choice. You may read the uncompleted work from MQ (Kafka), Akka persistence (RDS, Cassandra) etc.
SBR has open source solution. That is an advanced topic if your business logic works.
https://github.com/TanUkkii007/akka-cluster-custom-downing
The general outline of a solution is to use Akka Cluster, Cluster Sharding, and Akk Cluster Singleton. When the cluster is considered formed (generally when some minimum number of members have joined the cluster), you start the Cluster Sharding system (sharding work items by the DB's primary key) and then a Cluster Singleton will read the DB table and send work items to Cluster Sharding for distribution among the nodes of the cluster. Akka Streams and particularly Alpakka's Slick JDBC integration may prove useful within the singleton. Another cluster singleton to periodically check on jobs may also be useful to recover from cluster node failures (but see below for something to consider there).
Two notes:
If using Cluster Sharding and Cluster Singleton, you probably want to consider what happens in a split-brain situation: this is a distributed system and the probability of a split-brain eventually happening can be presumed to be 100%. In the split-brain scenario, you will very likely have the same jobs being performed simultaneously by different sides of the split, so you need to ask if that's acceptable in your use-case.
If not, then you will need a component which monitors the communications between nodes in your cluster to detect a split-brain and takes steps to resolve the condition: Lightbend's Split Brain Resolver is a good choice if you aren't interested in implementing this yourself.
In a related vein, if the jobs consist of many steps which must be performed, a question to ask is, if a cluster or node fails after completing, say, eight of ten steps, is it acceptable to redo steps 1-8 vs. starting with step 9? If the answer to this is "no", then you'll need to persist the intermediate state of the job. Akka Persistence is a great choice here, though you may want to read up on event sourcing. If using Persistence with Cluster Sharding and Cluster Singleton, it should be noted, you will almost certainly need to handle split-brains (see previous item).

Inject Jar and replace classes in running JVM

I want to be able to replace and add some classes to an already running JVM. I read that I need to use CreateRemoteThread, but I don't completely get it. I read this post on how to do it (Software RnD), but I can't figure out what it does and why. Besides that, it only introduces new classes, but doesn't change existing ones. How can I do it with C++?
You don't even need CreateRemoteThread - there is an official way to connect to remote JVM and replace loaded classes by using Attach API.
You need a Java Agent that calls Instrumentation.redefineClasses.
public static void agentmain(String args, Instrumentation instr) throws Exception {
Class oldClass = Class.forName("org.pkg.MyClass");
Path newFile = Paths.get("/path/to/MyClass.class");
byte[] newData = Files.readAllBytes(newFile);
instr.redefineClasses(new ClassDefinition(oldClass, newData));
}
You'll have to add MANIFEST.MF with Agent-Class attribute and pack the agent into a jar file.
Then use Dynamic Attach to inject the agent jar into the running VM (with process ID = pid).
import com.sun.tools.attach.VirtualMachine;
...
VirtualMachine vm = VirtualMachine.attach(pid);
try {
vm.loadAgent(agentJarPath, options);
} finally {
vm.detach();
}
A bit more details in the article.
If you insist on using C/C++ instead of Java API, you may look at my jattach utility.

Embedding Jetty 9.3 with modular XmlConfiguration

I am migrating from Jetty 8.1.17 to Jetty 9.3.9. Our application embeds Jetty. Previously we had a single XML configuration file jetty.xml which contained everything we needed.
I felt that with Jetty 9.3.9 it would be much nicer to use the modular approach that they suggest, so far I have jetty.xml, jetty-http.xml, jetty-https.xml and jetty-ssl.xml in my $JETTY_HOME/etc; these are pretty much copies of those from the 9.3.9 distribution. This seems to work well when I use start.jar but not through my own code which embeds Jetty.
Ideally I would like to be able to scan for any jetty xml files in the $JETTY_HOME/etc folder and load the configuration. However for embedded mode I have not found a way to do that without explicitly defining the order that those files should be loaded in, due to <ref id="x"/> dependencies between them etc.
My initial attempt is based on How can I programmatically start a jetty server with multiple configuration files? and looks like:
final List<Object> configuredObjects = new ArrayList();
XmlConfiguration last = null;
for(final Path confFile : configFiles) {
logger.info("[loading jetty configuration : {}]", confFile.toString());
try(final InputStream is = Files.newInputStream(confFile)) {
final XmlConfiguration configuration = new XmlConfiguration(is);
if (last != null) {
configuration.getIdMap().putAll(last.getIdMap());
}
configuredObjects.add(configuration.configure());
last = configuration;
}
}
Server server = null;
// For all objects created by XmlConfigurations, start them if they are lifecycles.
for (final Object configuredObject : configuredObjects) {
if(configuredObject instanceof Server) {
server = (Server)configuredObject;
}
if (configuredObject instanceof LifeCycle) {
final LifeCycle lc = (LifeCycle)configuredObject;
if (!lc.isRunning()) {
lc.start();
}
}
}
However, I get Exceptions at startup if jetty-https.xml is loaded before jetty-ssl.xml or if I place a reference in jetty.xml to an object from a sub-configuration jetty-blah.xml which has not been loaded first.
It seems to me like Jetty manages to do this okay itself when you call java -jar start.jar, so what am I missing to get Jetty to not care about what order the config files are parsed in?
Order is extremely important when loading the Jetty XML files.
That's the heart of what the entire start.jar and its module system is about, have an appropriate set of properties, the server classpath is sane, and ensuring proper load order of the XML.
Note: its not possible to have everything in ${jetty.home}/etc loaded at the same time, as you will get conflicts on alternate implementations of common technologies (something start.jar also manages for you)

How to integration test an object with database queries

How can i write unitintegration tests that talk to a database. e.g.:
public int GetAppLockCount(DbConnection connection)
{
string query :=
"SELECT"+CRLF+
" tl.resource_type AS ResourceType,"+CRLF+
" tl.resource_description AS ResourceName,"+CRLF+
" tl.request_session_id AS spid"+CRLF+
"FROM sys.dm_tran_locks tl"+CRLF+
"WHERE tl.resource_type = 'APPLICATION'"+CRLF+
"AND tl.resource_database_id = ("+CRLF+
" SELECT dbid"+CRLF+
" FROM master.dbo.sysprocesses"+CRLF+
" WHERE spid = ##spid)";
IRecordset rdr = Connection.Execute(query);
int nCount = 0;
while not rdr.EOF do
{
nCount := nCount+1;
rdr.Next;
}
return nCount;
}
In this case i am trying to exorcise this code of bugs (the IRecordset returns empty recordset).
[UnitTest]
void TestGetLockCountShouldAlwaysSucceed();
{
DbConnection conn = GetConnectionForUnit_IMean_IntegrationTest();
GetAppLockCount(conn);
CheckTrue(True, "We should reach here, whether there are app locks or not");
}
Now all i need is a way to connect to some database when running a unit integration testing.
Do people store connection strings somewhere for the test-runner to find? A .ini or .xml or .config file?
Note: Language/framework agnostic. The code intentionally contains elements from:
C#
Delphi
ADO.net
ADO
NUnit
DUnit
in order to drive that point home.
Now all i need is a way to connect to some database when running a unit integration testing.
Either use an existing database or an in-memory database. I've tried both an currently use an existing database that is splatted and rebuilt using Liquibase scripts in an ant file.
Advantages to in-memory - no dependencies on other applications.
Disadvantages - Not quite as real, can take time to start up.
Advantages to real database - Can be identical to the real world
Disadvantages - Requires access to a 3rd party machine. More work setting up a new user (i.e. create new database)
Do people store connection strings somewhere for the test-runner to find? A .ini or .xml or .config file?
Yeap. In C# I used a .config file, in java a .props file. With in-memory you can throw this into the version control as it will be the same for each person, with a real database running somewhere it will need to be different for each user.
You will also need to consider seed data. In Java I've used dbUnit in the past. Not the most readable, but works. Now I use a Ruby ActiveRecord task.
How do you start this? First can you rebuild your database? You need to be able to automate this before going to far down this road.
Next you should build up a blank local database for your tests. I go with one-per-developer, some other teams share but don't commit. In a .NET/MS SQL world I think in memory would be quite easy to do.